The Correspondence Method as a Data-Gathering Technique in Qualitative Enquiry

Introduction

As the use of correspondence as a primary data collection method in qualitative research is very new, little literature exists on its use to date. In this article, I attempt to fill that void by discussing the pros and cons of the correspondence method. Though the use (or usefulness) of letters as data appears acceptable within qualitative enquiry, they have traditionally been allocated a peripheral, secondary status, often used to verify or clarify findings from primary sources. Using letter writing as a qualitative data gathering strategy and as a primary data source, as reported in this article, is very unusual indeed. In fact, as far as this can be established, this article forms the first attempt to explore some issues and problems associated with undertaking this type of study.

In this article, the problems of representation of the self in a correspondence study of women who self-harm are discussed. This is followed by an exploration of the dilemma of ‘invisibility’ or becoming invisible within the research process, as participants do not meet in face-to-face interaction. Two questions are posed. First, in what ways does collecting data ‘invisibly’ through postal means, affect the research process? And second, to what extent can this method be viewed as an extension of ‘traditional’ qualitative research principles and practices? In reflecting on these issues it is concluded that more researchers should seriously consider using the correspondence method as a primary data-gathering technique since there are unique advantages that cannot be reproduced through other methods.

Background

Very few researchers have seriously considered the merits and demerits of undertaking research via correspondence. The reasons for this lack of consideration are unclear. Currently little information exists on the use of correspondence as qualitative data in health research, the exception being a seminal study by Kralik, Koch, and Brady (2000). In this study, Kralik and colleagues used a combination of e-mail and letter writing to collect life history data from a group of women (age ranging from 30 to 50 years) who were experiencing chronic illnesses. It appears that the core sample comprised 16 committed participants whose data informed the bulk of the study. Although Kralik et al. do not specify the number of participants who used only e-mail, it is implied that the majority of the ‘core group’ of 16 participants were in very frequent contact by e-mail, although ‘several’ used handwritten letters. Throughout their study, Kralik et al. corresponded with a large group of women (80 in total), although half of this number “wrote very intermittently” (2000, p. 911) and some others were reported to have “difficulty with reflection and self-disclosure” (2000, p. 911).

Kralik and colleagues (2000) concluded that there was no literature comparable to their study. They found that there had been very little debate on the use of e-mail for data collection apart from a very small amount that focuses on e-mail as a tool for concept synthesis (Bunting, Russell, & Gregory, 1998). Operationalising a new method caused both authors problems. Many of these problems (which are discussed below) surround issues that can be viewed as advantageous and disadvantageous to the study in certain circumstances or at particular junctures of the research process (see, for example, Kralik et al., 2000, p. 915).

Most of the honored standard qualitative texts do not mention the possibility of using letters as a main source of data. For example, discussion of this topic in Collecting and Interpreting Qualitative Materials (Denzin & Lincoln, 1998) is typical of the envisaged role of the letter within a qualitative study:

The authors later give the following brief description of the uses of correspondence:

Perhaps most telling of all is that despite the claim that letters are a “particularly interesting field text” (Clandinin & Connelly, 1998, p. 167), letters are accorded less than one page in a work of 462 pages on data collection and analysis (Denzin & Lincoln, 1998).

Texts primarily concerned with documentary analysis procedures are similarly silent on the use of letters as primary data sources. However, within a generalised discussion of the merits of ‘analysing documentary realities’ Atkinson and Coffey (1997) exhort:

To be clear, these authors are not advocating the use of correspondence as a data gathering strategy, as their primary focus is on exploration of the different forms of representation, functional and performative features of documentary analysis. Atkinson and Coffey (1997) are therefore joining other prominent qualitative authors in exhorting the use of documentary data, but leaving untouched all issues of practicality and strategy.

Many authors discuss these functional features of documentary analysis at length:

However, none of the above authors concentrate in any coherent form on either the use of correspondence as a primary data source, or indeed on methodological issues and dilemmas that arise in so doing.

Methods

The study that forms the basis for this article focused on women's self-harm (Harris, 2000). The decision to use the correspondence method grew from a failure to establish a meaningful face-to-face dialogue with women engaged in self-harm. The reasons for this failure were two-fold. First, this particular field is dominated by quantitative literature, which was initially mystifying. Later, it became obvious that the reasons were bound up with the medical model derivatives used by the psychiatrists and psychoanalysts who are the main spokespersons in the field. However, second, and more pertinent for my study, it also became apparent that the latter authors utilize quantitative data from hospital and clinic records because it is extremely problematic to obtain ‘live’ qualitative data from real respondents. This was caused partly by the nature of the phenomenon under study. Self-harm is an intensely stigmatized and embarrassing practice (Harris, 2000). People engaged in self-harm who present at emergency rooms for treatment, for the most part, want to receive that treatment and then depart as quickly as possible. They are not inclined to wait around to speak to researchers, or make appointments to do so. Neither do they respond to requests for interviews on posters in emergency rooms. In fact, the only time that a researcher might fortuitously come into contact with a person who self-harms would be at an emergency room, probably late at night and probably in an intensely critical situation. Face-to-face interviews would be difficult to arrange, harrowing for all participants and very probably highly unethical to undertake.

In compiling the literature review for the study, a pen pal network — a national organization for women who self-harm — was contacted. Correspondence ensued with the network coordinator, who eventually agreed to circulate details of the study, including a request to participate, among the membership. The request asked for ‘life stories’ in general and particularly details of contact with emergency room staff. Thus, the women ‘self-selected’ in terms of participation and, to a large extent, self-determined the content of their letters. The brief was deliberately left ‘wide’ so that the women would be able to express as much or as little as they wished on any topic at all concerning their lives. The study comprised correspondence between six women and myself over a period of twelve months. All participants were consenting adults (over 18 years of age) who had read the study outline and wrote of their own volition and it was therefore unnecessary to establish formal ‘informed consent‘, as this was implicit in their decision to participate.

The letters received were extremely detailed and covered all aspects of the women's lives. They were intensely personal accounts that detailed highly stigmatized practices (for example, removing stitches in the presence of emergency room staff and battles over whether it is necessary to see a psychiatrist following such incidents). The extremely informative texts provided in-depth accounts that seemed to transcend social barriers to description of stigmatized behaviours which may pertain in a face-to-face interview situation. For example, several women described in graphic detail incidents of childhood sexual abuse by family members which they consider to have been at least partially causative in their self-harming behaviour. It was also possible to gain insight into the ways that the women absorb societal reactions to self-harm. This was particularly evident in their portrayal of reactions to self-harm scars.

Although a method borne of necessity, correspondence proved a viable and productive means of gathering data in an otherwise intractable field.

Representation of the self

The most difficult aspect of the study involved working out how to respond to the letters; how to express some of myself but balance empathy with the need to steer away from counseling or advice giving; how to project an opinion without in any way judging the letter writer; and how to draw boundaries between being honestly interested in practices and acts and appearing salacious. The letters contained descriptions of intense critical life events (see above) and frequently, descriptions of gore. I wrote the following section in the research journal:

The letters demanded a very specific type of response that had to include exactly the right balance of ‘care’ (about the participant and what they were describing) and non-judgemental response. The response also had to contain exactly the right amount of empathy and the right amount of encouragement for them to want to continue with the study. This was unlike any other letter I had written before or have written since. For example, in most letters written to family members, it is usual to describe everyday events in a trivial fashion. In the self-harm study, it would be inappropriate to make trivial comments in a reply.

Over the study period, it was possible to observe that these ‘balances’ or adjustments in the reply letter had to be tailored correctly in direct proportion to the content of the life story. For example, the content of the letters was frequently dramatic, but the reply could not be similarly dramatic. All the same, it must demonstrate recognition — that the correspondent's statement had been heard, had registered, and had an effect.

It seems apparent on reflection that many of these dilemmas were engendered by the medium of the letter, which is an intensely private and personal document from one individual to another for a specific purpose. Put differently, we are used to writing letters to close friends and confidantes in this type of language — we use an entirely different style for business and other impersonal communication in the public domain. The letters written by the participants were in a personal, confidential style, as though to a friend or relative. The difficulty was that in fact, neither situation pertained. This particular hurdle was perhaps the most complicated methodological issue to solve in the entire study.

In the early days of the project, composing replies was extremely difficult. An entry from the research journal describes this dilemma: “Sat here for hours, trying to write back to S. Scrubbed words out over and over. Why is this so hard?” Thus, wording was carefully and painstakingly considered, hoping to strike just the right balance between these competing pressures of being honest, friendly, non-judgemental, encouraging, and non-counseling. Wondering whether the participants would continue to write and keep the information channels open engendered researcher anxiety. Effectively, it is imperative for the researcher to stay ‘on side’ — to remain receptive and empathic however emotive or graphic the content of letters. The content of the letters, as personal and private insights into people's lives, also invited partiality in responses. Descriptions of poor or oppressive health treatment engendered empathic anger. As the length of the project increased, it became progressively difficult to retain even a semblance of detachment from the data.

A further dilemma surrounded which of my ‘selves’ to show to the women? I could not show the ‘professional’ self, replying to their letters in the tone of a faceless bureaucrat, because of the emotional nature of the texts. These texts were full of intense chaotic emotion, but my own existence as an academic is comparatively ordered (some would say boring). Difference from one's respondents in important respects has been debated in the literature and it is well-recognised that ‘class’ and gender differences between respondent and interviewer may crucially affect the data collected (Sword, 1999, pp. 271–2). However, the literature did not offer guidance on the particular dilemma faced by this researcher. In a pragmatic attempt to solve this dilemma, the decision was made to reply at a personal level, showing as much humane concern as was possible (but not venturing towards becoming patronising or judgemental). Direct comment on the gory practices was not made, but the correspondent was given to understand that she had been heard and that the words she had written had had an effect on both researcher and the study as a whole. In attempting to ‘solve’ the problem of self-presentation and demonstrations of empathy, parallels were drawn with difficult periods of the researcher's own life that in some way mirrored the intensity of that which was described by the correspondent. For example, one reply stated:

The correspondents differed in the amount they wished to know about my personal situation, and each of these disclosures on my part was different in both tone and amount. The response above was directly tied to a letter in which the correspondent had disclosed both graphic detail of cutting episodes and sexual abuse as a child. Though there are certainly no hard and fast rules that could be applied here, one of the ‘balances’ that had to be made was an acute need to disclose at least some personal information in situations where correspondents had sent copious descriptions of highly personal and traumatic events. The art was to tailor the response directly to the letter that preceded it.

Researching or Advising?

Kralik and colleagues (2000) claim that their respondents asked for advice but they resisted the temptation to dispense it. None of the correspondents in my study specifically and directly asked for advice on their situation. This was simultaneously a relief and a matter of concern. However, it became a considerable source of personal dissatisfaction that I was party to the harrowing details of the women's lives and yet, in nearly all important respects, ineffectual in assisting them toward a resolution. This was a curiosity as it had not occurred in research situations where the medium of contact was face-to-face. In analysing this situation, it seemed that the underlying and contributory problem to the evidential feeling of dissatisfaction was the issue of reciprocity. The correspondents had given me their life stories freely, which had then been used to compile research reports, but nothing had been given back. It must be noted that this was a dilemma affecting purely the researcher, as several women wrote toward the end of the study that they felt writing their life stories had enabled them to achieve a resolution:

It would obviously not have been appropriate to offer correspondence counseling and neither myself nor the correspondents wished for this. However, striving for empathy in situations which are far from your own reality is not easy. For example, one woman wrote a harrowing account of an encounter with a psychiatrist and how very difficult it had been for her to ‘be strong’ under pressure. My reply included the following extract:

Reviewing this response in retrospect, the overtly feminist gloss and the vehemence with which the passage was written are evident. There is an effort to abstract from the correspondent's position to the researcher's own, without too much detailed explication of the parallels between the situations. There is a very visible attempt to make a claim to joint oppression and sisterhood toward the end of the passage and this was clearly part of the attempt to identify correctly with the woman's situation and provide an empathic response that was neither judgemental of her behaviour nor directed toward offering a resolution, both of which would have been clearly inappropriate.

‘Invisibility’

There are a number of unique problems brought up by the correspondence method which stem from the fact that the participants do not meet face-to-face. One of the most frustrating aspects of the correspondence method concerns the ‘one step removed’ aspects of this means of data collection. For example, in the self-harm study, women often wrote highly charged emotional letters that appear to have been painful to write (many were tear-stained). On receipt of these letters, consideration was given to whether an interview would have continued in a face-to-face situation. Mostly likely in that situation I would have offered to break and make tea or come back later if she so wished. It was therefore difficult to observe emotional portrayals on paper as, in many crucial respects, the humane elements of research (expressing concern, empathy, and consideration for participants) were impossible to put into action. It should be noted also that these problems, and the considerable unease that they generate, increase as the project continues since the researcher becomes more closely involved with the minutiae of participants' lives.

A further issue that is intrinsic to the correspondence method is the lack of ‘immediacy’ between question and response. There was usually very little delay in responding to any specific question that I asked, but even so, using postal services meant that between three and seven days elapsed. ‘Time-lag’ can be a frustrating aspect of the method but may possibly have beneficial effects for participants:

Reflecting on the data in this way helped this participant to ensure she had made her account comprehensive and perhaps, to adjust the wording where she felt this was necessary either for clarity or emphasis. One detrimental effect of the ‘time-lag’ however, was that some participants ‘forgot’ to answer certain questions and it seems likely that this was caused by the lack of immediacy between receipt and reply. These situations engendered another dilemma caused by invisibility. Had the participant not answered this question because it was too painful or had she simply forgotten it was in my letter? If the former was the case, then to ask it again would be unethical, or at least insensitive and upsetting to the participant. In wrestling with how to react in these situations, the researcher again resorted to comparing this to face-to-face interviews. Most commonly, when interview participants ‘forget’ to answer a question it is because they are more interested in a tangential subject and it is fine to (respectfully) rephrase the original question to see if they had anything to add. In most cases where this occurred in the correspondence, the (rephrased) question was included in the next letter and nearly always received replies second time around.

These issues, which some may consider to be matters of research ethics, must however be balanced by a consideration of possible beneficial effects, both for the research process and for participants. One means of dealing with the unease of receiving a letter full of pain and the resultant distress was to consider whether, if we had been face-to-face, the woman would have been able to express her feelings so completely and directly. Many of the women described the cathartic effect of writing the letters and the researcher considered whether for example, keeping a personal diary would achieve the same effect. However, several women stated that it was important to them that ‘someone saw this’ — that I was reading, considering and structuring the ongoing research around their texts. It seems important therefore, not to overlook this effect as merely a routine part of research practice.

Qualitative research principles and practice

In attempting to explore the merits and demerits of using the correspondence method as a means of gathering qualitative data, it seems apparent that the method does not naturally ‘fit’ within our understanding of the usual principles and practices of qualitative methodology. However, the extent to which using the correspondence method as a primary data-gathering technique appears to stand outside the established parameters of the qualitative research tradition may be gauged more accurately by reflecting on the purposes and outputs of research. This is demonstrated, for example, through comparing the correspondence method and face-to-face interviewing. The purpose of conducting a face-to-face interview might be to gain access to that person's social world, or to at least acquire a representation of their social world. The output would be (usually) a transcribed text, perhaps of their life history. At a functional level, there is no difference between this text and that supplied by the correspondence-study participants who each sent their life story and answered subsequent questions that these stories raised. The only major difference between the two scenarios is in the ‘immediacy’ of data gathered in the traditional ethnography and the ‘humane’ aspects of caring for research participants (discussed above). Indeed, it could be argued that the facility for reflection noted by correspondence study participants, may render the method superior to face-to-face interviewing in terms of accuracy of description.

The ‘invisibility’ of the correspondence method also posed restrictions to observing participants' emotions. However, failure to engage participants for face-to-face interviews pointed to the issue of the intensely emotional nature of the subject under study and this was proved conversely to be a bonus for the correspondence method. It is therefore entirely possible that the correspondence method works well because it allows for the expression and diffusion of intense emotions. It could be that it is far easier and less painful to write about intense emotional experiences than it is to explain them face-to-face with a researcher. This aspect of the correspondence method should not be overlooked, as there are many other intensely private and personal areas of the social world that may prove to be more accessible through this method than via interviewing techniques.

Conclusion

The use of correspondence as a data collection method has historically been relegated to a second class status, with ‘live’ material, collected at first hand in interview situations taking precedence. In this mode, the role of ‘texts’ (files, reports, official records) has been to support or verify accounts contained in live data. Correspondence as a primary data collection method, as used here, is extremely unusual.

Correspondence was used in a ‘difficult’ field (self-harm) that was not amenable to face-to-face interviewing because of the intense social stigma associated with these behaviours and the lack of opportunities to recruit participants in any of the usual ways. This fortuitous discovery points to the fact that the correspondence method may prove to be useful in other equally stigmatising or intensely personal circumstances. Thus, the issue of ‘invisibility’ within the research process (as participants do not meet in face-to-face interaction) affects the research data in both beneficial and detrimental ways. Whereas the major benefits to be gained involve circumventing the embarrassment that participants may otherwise experience in describing stigmatised practices, major detriments may be experienced by the researcher who, in composing replies, faces the dilemma of ‘balancing’ correct amounts of empathy with interest in the participants that stops short of salaciousness. Further, ‘balancing’ work must be undertaken to determine how to interact with complete strangers on a personal level through the very personal medium of letter-writing and how to represent the self in this context. This article charts the pragmatic solutions to these dilemmas. More studies of this kind are required to explore these issues in more detail. More researchers should seriously consider using the correspondence method as a primary data-gathering technique, as there are unique advantages that cannot be reproduced through other methods.